Blood Pressure With a Click of a Camera?

In this issue of the journal, Luo et al 1 investigate the feasibility of noncontact blood pressure (BP) measurement using a smartphone camera. The potential of this concept is fascinating. Each time a person uses a smartphone, an application controlling the camera could continually search for opportunistic moments (eg, when the person is still) to make BP measurements. Such passive and frequent BP monitoring during daily life with devices that are already in the pockets of many could help improve on the currently low hypertension awareness and control rates around the world.2 Noncontact cardiovascular monitoring with basic video cameras is actually not new. The measurement principle is the same as reflectance-mode photo-plethysmography (PPG, which is employed by pulse oximeters) except that ambient light rather than a dedicated light source serves as the illuminator. The external light is mainly absorbed by melanin in skin and hemoglobin in cutaneous blood vessels, and the residual light is reflected back to the camera. Each recorded video pixel of the skin, therefore, includes subtle color changes superimposed on the average skin color that are inversely related to the pulsatile arterial blood volume. Video PPG waveforms at various skin locations may then be extracted from the image sequence by way of signal processing. Early investigators focused on noncontact measurement of heart rate with video cameras,3 and the Philips Vital Signs Camera application for heart rate monitoring was introduced a few years later in 2011.4 A natural extension to this work, which has clinical application, is noncontact screening of atrial fibrillation via detection of irregularly, irregular pulse intervals in the video PPG waveforms.5 Although video camera BP measurement is less obvious and would seemingly require a leap of faith, previous investigators envisioned such an application via the popular pulse transit time principle6 and particularly the time delay between video PPG waveforms from the face and hand or 2 locations on the face.7–9 Luo et al1 may have conducted the largest study to date on noncontact BP measurement with a video camera. These investigators specifically used a standard smartphone front camera to extract video PPG waveforms from 17 facial locations and a volume-clamp, finger-cuff device to measure reference BP from of 1328 normotensive subjects. The investigators then applied machine learning techniques to 85% of these data to identify video PPG waveform features that best correlate with the reference BP measurements and develop a formula to convert these features to systolic and diastolic BP values in units of mm Hg (ie, cuff calibration). They finally tested the developed method on the remaining 15% of the data while blinded to the reference BP measurements. Their most intriguing finding is that the method was able to estimate systolic BP better than a model based on basic 10.1161/CIRCIMAGING.119.009531 2019 20